04.08.24
I'm interested in making new genre's of music using gen AI but so far I have found their genre interpolation capabilities very poor.
music generated using suno.com & udio.com. Combined with text-to-speech and words taken from this amazing blog article
If you reverse the audio you can bypass the copyright protection filter. Managed to smuggle in some Beatles - a day in the life sample and get suno to extend john lennons voice using the lyrics functionality.
Combined into a 2 hour mix using bitwig.
As a rule of thumb the output from genAI music generators sounds like awful stock music but if you keep experimenting and tweaking the prompt you can get the occasional interesting artefact. That's been my approach and I tend to throw away 90% of what I generate.